Investing Choices and Risk Measures

(Welch, Chapter 08)

Oleg Shibanov
New Economic School

February, 2014

Did you bring your calculator? Did you read the chapter ahead of time?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

1 / 35

Maintained Assumptions
In this part (starting from the previous chapter), we maintain the same
assumptions:
I

We assume perfect markets, so we assume four market features:

1.
2.
3.
4.

I
I

No
No
No
No

dierences in opinion.
taxes.
transaction costs.
big sellers/buyerswe have innitely many clones that can buy or sell.

We already allow for unequal rates of returns in each period.

We already allow for uncertainty. So, we do not know in advance what the
rates of return on every project are.

But in contrast to Chapters 6, we no longer assume risk-neutrality. We are

allowing for risk aversion now.

In this chapter, we lay the groundwork for understanding how investors

choose among many dierent projects.
You need this [a] to think as a manager about your companys investment
risk; [b] more importantly to think about your opportunity cost of capital,
E (r ).

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

2 / 35

OMIT: What Preference Assumptions Buy Us

What do we need to compare stocks if we can only invest in one?
State 1
5%

State 2
5%

State 3
15%

6%

4%

10%

Stock B2

6%

6%

5%

Stock B3

10%

5%

20%

Stock B4

10%

6%

20%

Base
Stock A
Stock B1

I
I

I
I

same, but 1% less in all

states
better in state 2, worse
in 1 and 3
same mean, higher variance
higher mean, higher
variance

Everyone prefers A to B1.

Not everyone prefers A to B2you must assume that one cares about states
equally to prefer A. (Simplest example: [a] you are only alive in state 2; or [b]
the probability of state 2 is 99%.)
Risk-averse investors would prefer A to B3.
For B4, we need to specify how we want to trade o risk vs. return.
Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

3 / 35

Some Particular Investments

1/4
1/4
1/4
1/4

Yellow
Red
Green
Blue

A
4
4
+6
+6

B
1
+9
+9
1

C
1.25
+1.25
+3.75
+1.25

D
+3
+13
+3
7

We will be using these four assets (A to D) in these slides. You can think of these
as rates of return (or, if you prefer, as dollar returns instead of rates of return).

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

4 / 35

What are the rewards of your above investment

opportunities?

Means are 1, 4, 1.25, and 3.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

5 / 35

What are the risks of your above investment opportunities?

Variances are
25, 25, 3.125, 50
Standard deviations are
5, 5, 1.77, 7.07

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

6 / 35

Population vs. Sample Statistics

If these returns are just representative historical realizations from a

population, you would divide by 3, not 4 in your computation of the variance.
In real life, we never have population statistics. We only have historical
sample statistics. And we really should not trust the historical statistics
eitherbut we do because this is the best alternative.

The standard deviation as a meaningful measure of risk applies only to your

overall portfolio. You do not care about the standard deviation of your individual
securities.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

7 / 35

Complete the table

Yellow
Red
Green
Blue
Mean (E )
Var
Sdv

A
4
4
+6
+6
1
25
5

Oleg Shibanov (NES)

B
1
+9
+9
1
4
25
5

C
1.25
+1.25
+3.75
+1.25
1.25
3.125
1.77

D
+3
+13
+3
7
3
50
7.07

AA
5
5
+5
+5

Investing Choices and Risk Measures

B B
5
+5
+5
5

C C
2.5
0.0
2.5
0.0

D D
0
10
0
10

February, 2014

8 / 35

Whats the risk of an (equal-weighted) ptf of A and B?

(HINT: First compute the rates of returns of the combination portfolio in each state.)

The portfolio returns of (A,B)=(0.5,0.5) are

2.5, +2.5, 7.5, 2.5
Demean to get
+25, 0, 25, 0
Sum, and divide by 4,
Var = 12.5

Sdv 3.5

A and B are uncorrelated, so 1/ 2 5% would have worked, too.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

9 / 35

Is the average portfolio or are the individual components

riskier? Why?

Dwell on how this 3.5% is lower than each individual Sdv of 5%.
The portfolio is safer than its components.
The reason is non-synchronicity.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

10 / 35

What kind of portfolio would youa smart investorhold?

Probably almost everything that you can get a hold ofalthough many stocks in
smaller proportions particularly if these stocks do not oer reasonable (high)
means.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

11 / 35

In real life, what portfolios should smart investors with

risk-aversion hold?

Something heavily diversied, perhaps reasonably close the market portfolio.

You may exploit anomalies, i.e., situations in which you think some stocks have
higher mean returns than they should have. (Remember factors from lecture 7).

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

12 / 35

What is your portfolio risk if you add C to your portfolio

vs. if you add D to your portfolio?
Yellow
Red
Green
Blue
Mean (E)
Var
Sdv

A
4
4
+6
+6
1
25
5

Oleg Shibanov (NES)

B
1
+9
+9
1
4
25
5

C
1.25
+1.25
+3.75
+1.25
1.25
3.125
1.77

D
+3
+13
+3
7
3
50
7.07

Half A, half C

Investing Choices and Risk Measures

Half A, half D

February, 2014

13 / 35

Is C or D the riskier investment in itself?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

14 / 35

If you already own A, is C or D the riskier addition?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

15 / 35

Why?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

16 / 35

If investors are smart, what is their A?

A heavily diversied portfolio like the market.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

17 / 35

If you are selling to smart investors either C or D, for

which of these two projects do you think will investors
clamor to invest in your project (i.e., accept a lower
expected rate of return)?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

18 / 35

The fundamental insight of investments

Investors (should) care about overall portfolio risk, not about the constituent
component risk.

From a corporate managerial perspective, it is not your projects that are low
risk in themselves that are highly desirable for your investors, but projects
which wiggle opposite to the rest of their portfolios

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

19 / 35

What is the synchronicity (correlation or beta) of our

project that may matter to what our investors like?
Our investors (or we) are interested in non-synchronicity with respect to what our
investors (we) are already holding.
In the CAPM, our investors are holding the market. Therefore, they want the
market-beta to be as low as possible.

From now on, consider A to be our market portfolio that our investors are
already holding.

We need a measure of how synchronous or non-synchronous any new

stock/asset/project is with respect to this portfolio.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

20 / 35

Reminder

Note: M(arket)=A.

Yellow
Red
Green
Blue
Mean (E )
Var
Sdv

M
4
4
+6
+6
1
25
5

Oleg Shibanov (NES)

B
1
+9
+9
1
4
25
5

C
1.25
+1.25
+3.75
+1.25
1.25
3.125
1.77

D
+3
+13
+3
7
3
50
7.07

M M
5
5
+5
+5

B B
5
+5
+5
5

C C
2.5
0.0
2.5
0.0

D D
0
10
0
10

25
5

25
5

3.125
1.77

50
7.07

Investing Choices and Risk Measures

February, 2014

21 / 35

Which is the best measure of risk?

Covariance has uninterpretable units.

Correlation has a scale problem.

I

The correlation would tell us that a security with rates of return

R = (0.9, 1.0, 1.1, 1.0) has the same 70.7% correlation with A as
S = 1000 R 1000 = (100, 0, +100, 0) doesbut $100 of R will clearly
contribute less risk to our portfolio M than $100 of S.

Therefore, we prefer measuring risk contribution of B or C by its market-beta

with respect to A (here = M).

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

22 / 35

Which is the best measure of risk II?

I

Beta is similar to correlation. It always has the same sign.

Beta can be interpreted as a slope. Put A (M) on the X axis, and your
project B (or C) on the Y axis. A slope of 1 is a diagonal line. A slope of 0 is
a horizontal line. A slope of is a vertical line.

Without alpha, beta tells you how an x% higher rate of return (than normal)
in the market will likely reect itself simultaneously in a i x% higher rate of
return in your stock.

Together with alpha, beta can be interpreted as giving you the best
conditional forecast of your projects rate of return, given a market outcome
scenarios rate of return.

Practical estimation of future market-beta from historical stock return data is

discussed in the book.

Yahoo!Finance lists estimates of betas for many stocks, too.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

23 / 35

What is the market-beta of the market (the S&P500)?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

24 / 35

Ceteris paribus, should/do investors prefer securities with a

higher beta or a lower beta with respect to their (market)
portfolio?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

25 / 35

Should/do high beta or low beta projects have to oer

higher average rates of return?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

26 / 35

Should/do high variance or low variance projects have to

oer higher average rates of return?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

27 / 35

If you own a rm consisting of $4 million invested in

Division C, and $6 million in Division D, what are this
rms returns?
M MM
C
D Firm Fm Fm
Yellow
4
5 1.25 +3
Red
4
5 +1.25 +13
Green
+6
+5 +3.75 +3
Blue
+6
+5 +1.25 7
Mean (E )
1
0
1.25
3
Var
25
25 3.125
50
i,M
1
0.25
1

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

28 / 35

Using the markets and your rms rates of return, what is

the market beta of your rm?

covar =

[(5) (1) + (5) (+6) + (+5) (+1) + (+5) (6)]

= 12.5.
4
betaF ,M = 12.5/25 = 0.5.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

29 / 35

Is there a quicker way to compute the overall market-beta

of your rm, based on its projects?

[Recall that C = 0.25; D = 1]

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

30 / 35

What statistics can you value-average? What statistics

can you not value-average.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

31 / 35

Hugely Important But Omitted

The mean-variance ecient frontier (almost equivalently, the mean-standard

deviation ecient frontier).
I

Covering it would require 1-2 full lectures. Though omitted, the

mean-variance ecient frontier is extremely important. It is the basis for
modern nance and for the CAPM.

The frontier gives you the optimal set of assets that you should hold if you
want to tolerate a risk of x%. It also tells you what expected rate of return
this portfolio (for your specic risk-tolerance) should give you.

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

32 / 35

The Sum of Variances

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

33 / 35

What is the correlation of stocks rates of returns from one

day to the next day?

Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

34 / 35

Annualizing Variance
There is an extremely important application of the variance-additivity formula:
I

Rates of return over time are usually uncorrelated (or you could use past
stock returns to outpredict future stock returns). Algebraically,
Cov (Rt , Rt+i ) 0
where the subscripts t and t + i refer to time periods, not to stocks (as our
subscripts usually do).

Lets presume that the per-unit-of-time standard deviation is constant. Lets

just call this number .

In this case, the following approximation is not bad:

Sdv (R0,T ) T

This annualized sd is also used in the Sharpe-ratio, a (badly awed but

common) measure of investment performance that divides the historical
average rate of return (net of the risk-free rate) by its standard deviation.
(The SR of a portfolio grows with the square-root of time.)
Oleg Shibanov (NES)

Investing Choices and Risk Measures

February, 2014

35 / 35